As the "steam engine of the Fourth Industrial Revolution," generative AI applications have now covered the entire lifecycle of the manufacturing industry, bringing significant business value in areas such as product R&D and design, manufacturing operations, supply chain, marketing and sales, intelligent customer service, and knowledge bases. Meanwhile, with the rise of the generative AI wave, "AI+" appeared for the first time in this year's government work report at the Two Sessions, emphasizing the promotion of deep integration between AI and large-scale models and key areas of the real economy, building digital industrial clusters, and further accelerating the digital transformation of the manufacturing industry.
According to a recent report by the Boston Consulting Group, manufacturing executives generally rank artificial intelligence (including generative AI) as one of the top technologies likely to positively impact their operations and believe it can deliver a potential return on investment. Meanwhile, MarketResearch predicts that the global market size for generative AI technologies in manufacturing will grow from $223 million in 2022 to approximately $6.399 billion in 2032, representing a compound annual growth rate of 41.06%.
Leveraging Generative AI with High-Quality Data as a Foundation
The manufacturing industry generates massive amounts of data rapidly and continuously across all stages, including production, quality control, and management. It is estimated that the manufacturing sector produces approximately 1,812 PB of data annually, exceeding the data generated by industries such as telecommunications, finance, and retail. However, despite its rich application scenarios, the manufacturing industry often faces isolated and unconnected data sources. This makes it difficult for basic models to obtain high-quality, economical, secure, structured, and easily accessible datasets, resulting in the predicament of massive amounts of data becoming "islands."
Generative AI requires massive amounts of data to create foundational models; therefore, establishing a high-quality, end-to-end data foundation is a powerful driving force for the rapid implementation of generative AI technology in manufacturing. Amazon Web Services (AWS) provides end-to-end data strategies and services, covering every stage from data ingestion, storage, and retrieval, databases, data lakes, to data analytics, business intelligence (BI), and data governance, and further to AI and machine learning innovation. This helps enterprises fully unlock the value and potential of their data, empowering the next phase of scenario-based AI technology applications. Furthermore, a key aspect of using data as the core of business decisions is the ability for enterprises to connect all their data, regardless of where it is stored. AWS is driving a "Zero-ETL" future, allowing customers to easily integrate data from across their systems without having to extract, transform, and load (ETL) data between different services, enabling more efficient and informed data-driven decision-making.
With the support of Amazon Web Services and cloud-native services, Siemens focused on a design framework centered on microservices and event-driven architecture. By integrating various disparate data sources, Siemens achieved integrated data management and created the "Dayu" data platform. With the rise of generative AI technology, Siemens, building on the robust data foundation established by the Dayu team, completed the flexible construction of "Xiaoyu," an intelligent knowledge base and intelligent conversational robot based on its own model, in just three months. "Xiaoyu" possesses core capabilities such as natural language processing and knowledge base retrieval, significantly improving the efficiency of information acquisition for internal employees. In the words of Siemens Group's head of IT data analytics and artificial intelligence, "Without 'Dayu,' there would be no 'Xiaoyu,'" demonstrating how a strong data foundation can greatly enhance the implementation of generative AI scenarios.
Application is king: identify the core application scenarios.
With a high-quality data foundation, the next step for enterprises is to start from practical applications and solutions, organically integrate digital technology with core business, and quickly deploy generative AI to extract value. For B2B manufacturing customers, building application solutions requires inference, and a significant unavoidable issue is the high cost of inference. Therefore, manufacturing enterprises urgently need to focus on return on investment, using "application is king" as the standard, finding a balance between model accuracy and inference cost, and achieving an ideal return on investment while solving business challenges.
Amazon Web Services (AWS) is committed to driving the growth path of manufacturing through generative AI. By lowering the barriers to building critical pathways for generative AI applications, AWS fully penetrates the manufacturing value chain. Working closely with customers to deeply understand their pain points and needs, AWS, together with its partners, develops customized solutions for specific business scenarios in manufacturing, such as industrial product design and marketing material generation that heavily rely on human intervention, and complex internal and external data collection. These solutions enable manufacturing companies to fully realize the potential of generative AI.
In the field of industrial product design, traditional industrial concept design is generally done by hand-drawing. This presents challenges such as long design cycles, the need to balance the workload of designers with rapidly growing business demands, and fluctuations in design quality due to differences in staff skill levels and staff turnover. These factors combined result in high labor costs, low efficiency in concept output, and low concept approval rates during the concept design phase. Amazon Web Services (AWS) and its partners have jointly developed a generative AI solution that enables rapid concept prototyping through text-to-image and image-to-image generation methods. This solution can generate multiple improved design drafts at once, allowing clients to select the optimal option. Furthermore, clients can subsequently make targeted adjustments and optimizations to the selected materials and submit integrated renderings with a single click, effectively simplifying the process, reducing concept design costs, and accelerating overall industrial design efficiency.
Amazon Web Services partner Nolibox, leveraging Amazon Web Services' services, helped Haier Innovation Design Center create the nation's first generative AI industrial design solution through text-to-image and image-to-image generation. This solution accelerated the overall concept design process at Haier Innovation Design Center by 83%.
In the field of enterprise knowledge bases, according to data provided by Capgemini, 80% of enterprise data is currently unstructured (documents, help website support documents, etc.). Due to the continuous surge in data and its often dispersed nature, employees frequently face challenges such as inaccurate content and difficulty in finding key information when searching for crucial data. Amazon Web Services, together with its partners, leverages generative AI technology to build enterprise-grade intelligent knowledge bases for clients. By integrating search engines and large language models, it helps clients quickly build knowledge base dialogue applications through intelligent knowledge base architecture design, large language model pre-training, and artificial intelligence and machine learning technologies. This enables employees to quickly find accurate and timely content within the knowledge base, transforming raw enterprise data into usable digital assets, significantly improving production and office efficiency, and enhancing customer experience.
A leading global home appliance customer faced significant pressure in its after-sales service team, lacking a multilingual global knowledge base and limited intelligent question-and-answer generation capabilities. Amazon Web Services partner Hongyi Technology leveraged Amazon Web Services' solution guide, which combined knowledge bases, search engines, and large language models, to help the customer quickly build a precise, searchable, and question-and-answer-enabled enterprise knowledge base. This enabled the customer to improve their global after-sales service experience while transforming raw corporate data into usable digital assets.
The collaborative development of large and small models helps customers achieve the "last three kilometers" of application deployment.
The manufacturing sector is highly fragmented, with significant industry knowledge barriers. Therefore, for specific sub-sectors within manufacturing, it's difficult to find sufficient publicly available core process data for pre-training large models, while general-purpose large models cannot meet the customized needs of vertical scenarios. Thus, manufacturing companies don't need to blindly pursue the largest possible models. For specific tasks and vertical scenarios, smaller models can help companies achieve efficient computation and inference within limited resources. Currently, the coexistence of large and small models remains a major trend for the foreseeable future.
Furthermore, the "last three kilometers" of generative AI application deployment are crucial, requiring significant engineering resources and investment, including cloud computing infrastructure, data engineering, model tuning, and user interface development. In the early stages of generative AI application, Amazon Web Services (AWS) adheres to the philosophy and practice of "helping them get started and supporting them along the way," leveraging its rich professional technical support resources, including prototype development teams, data scientists, industry architects, and professional service teams, to work with numerous partners to truly help customers solve the "last three kilometers" of application deployment.
Innovation and continuous empowerment of the manufacturing industry to deepen innovation and transformation
In an era where generative AI accelerates innovation, Amazon Web Services (AWS) continues to launch new solutions and technological tools, empowering customers to easily build and scale generative AI technologies. At re:Invent 2023 last December, AWS unveiled Amazon Q—a new generative AI-enabled assistant specifically designed to meet the needs of office environments and can be customized to meet individual business requirements. Whether building on AWS, using internal data and systems, or using AWS applications for business intelligence (BI), contact centers, and supply chain management, Amazon Q is a powerful generative AI-based assistant that helps businesses of all sizes and across all industries securely utilize generative AI.
In the manufacturing industry, building a comprehensive service system, including intelligent customer service, to enhance the customer experience is crucial for business success. However, currently, 70% of equipment manufacturing companies lack after-sales systems. Amazon Q, part of Amazon Connect, has been officially released. Amazon Connect is a cloud contact center that enables businesses of all sizes to deliver superior customer experiences at a lower cost. Amazon Q in Amazon Connect detects customer issues based on real-time conversations between customers and customer service representatives, automatically responding, providing suggestions, and offering relevant information. This improves customer satisfaction while reducing time and costs associated with customer service staff training and problem-solving.
Furthermore, Amazon Web Services (AWS) continues to expand and enrich the models available on Amazon Bedrock, enhancing the ability of manufacturing enterprises of all sizes to quickly test, build, and deploy generative AI applications within their organizations. For example, Anthropic's leading models Claude 3 Sonnet and Claude 3 Haiku, as well as Mistral AI's two high-performance models, Mistral 7B and Mixtral 8x7B, have recently become officially available on Amazon Bedrock, increasing users' freedom to choose high-performance base models on Amazon Bedrock. Simultaneously, AWS also announced a strengthened collaboration with NVIDIA, with NVIDIA's next-generation Blackwell GPU platform soon to be available on AWS, aiming to provide customers with secure, advanced infrastructure, software, and services to help them unlock the next generation of generative AI capabilities.
From the steam engine and electricity to information technology, each industrial revolution has brought about tremendous technological progress and industrial transformation, profoundly impacting human society. Today, with the rapid development and widespread application of generative AI, a new wave of technology represented by artificial intelligence has become a significant driving force for the Fourth Industrial Revolution. Amazon Web Services will collaborate with hundreds of thousands of partners worldwide to jointly develop generative AI solutions for key manufacturing scenarios, helping enterprises leverage generative AI to create new growth engines, and fully promoting the deep integration of "AI+" with manufacturing, thus fostering the high-end, intelligent, and green development of the manufacturing industry.
